Automation
Cases.

Monthly financial reconciliation used to mean hours of manual work: pulling transactions one by one, converting each at the correct period exchange rate, assembling a readable summary by hand. Previous attempts to solve this in Make and n8n failed. Built in Relay.app in a single working session.

The workflow runs on a fixed schedule: on the first of each month it pulls all PayPal transactions for the previous period, fetches the National Bank of Georgia exchange rate via their public API, applies the correct conversion to each transaction, and assembles a structured financial summary — automatically delivered by email. No manual input. No open tabs. No forgotten months.

Same task. Zero effort after setup.

A three-workflow system that removes email from the meeting coordination loop entirely. New booking, cancellation, or reschedule — each triggers an instant Telegram notification with client name, meeting time, and direct link. No inbox required.

Two additional scheduled workflows handle the reminder layer: a structured alert 24 hours before each call, and a final check-in 1 hour before. The architecture uses conditional branching to handle all three event types through a single webhook entry point — clean, maintainable, no duplicate logic.

First real meeting confirmed the system: notification arrived before the confirmation email did.

A new document appears in a monitored Google Drive folder. The workflow picks it up, reads the content, and uses AI to identify and structure discrete tasks — parsing intent, not just formatting. Each task becomes a Trello card: title, context, and description populated automatically. No copy-paste, no card-by-card manual entry, no formatting convention required.

When a card moves to "Ready to start", the second layer activates: a Gmail draft is generated with the specific task details, a direct link to the card, and a ready-to-send team notification. Pre-filled, contextually accurate, one click to send.

One document drop triggers the full chain: backlog populated, cards structured by AI, team coordination queued. The overhead disappears.

Email is where critical information goes to be forgotten. This workflow changes the routing: specific senders, subject patterns, or keyword triggers in Gmail activate an immediate Telegram notification — structured, readable, actionable — without opening the inbox.

Designed for invoices, payment reminders, tax notices, and time-sensitive operational messages. The notification includes sender, subject, and key content extracted from the message body. Configured once against the sources that matter; runs silently in the background.

Built for anyone who checks Telegram before email — which is most people.

Two input streams, one analysis layer. The first runs on schedule: the workflow pulls fresh postings from a connected set of European job boards, filters by role and location, and queues relevant results for analysis. The second accepts manual input — forward any vacancy link or text directly to a Telegram bot and it enters the same pipeline immediately.

AI reads each posting against a pre-configured candidate profile: salary floor, format preference, seniority level, domain fit, stack tolerance, location constraints. A parallel red flag scan runs on every posting — vague scope masking multiple roles, missing company identity, payment conditions that don't hold up, probation terms written to benefit one side, requirements designed for a unicorn at a junior budget.

Output arrives in Telegram: structured match assessment, breakdown of what fits and what doesn't, flags ranked by severity. Thirty seconds to a decision that used to take fifteen minutes of careful reading.